Editorial illustration for xAI used Anthropic’s Claude via personal accounts after access revoked for months
xAI used Anthropic’s Claude via personal accounts after...
Elon Musk’s xAI trained its models by borrowing from a rival long after the rival told it to stop. The technical term is data mooching. For months after Anthropic blocked official access in January, xAI engineers logged into Claude using personal accounts and a third-party service. This is the Silicon Valley equivalent of sneaking into the neighbor's yard to use the hose after they turned off the water.
The story isn't the borrowing. Everyone does that. It's what happened next.
The core team for building models fell below five people. Key leaders quit. Someone deleted weeks of work by accident.
These aren't isolated failures. They signal a project that's coming apart at the seams. Musk is now leasing his own company's computing infrastructure to Anthropic and Google, the very firms he claims to be racing against.
It’s a strange business: building a car while renting your factory to other automakers. The move suggests his own model training pipeline is stalled. Renting out your servers to your competitors is not a pivot.
It's a retreat.
Related Reading
Browse all Anthropic news →
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google tests visual 'magazine-style' UI for Gemini 3 Pro users
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
Related Topic Trump cracks down on Anthropic after Amazon tip; staff largely foreign
Open Source
Related Topic Claude Mythos highlights EU AI safety gaps, says researcher Caroli
Policy & Regulation
From Archives Study examines temporal preference concepts in large language models
From Archives Errorquake-10k Benchmark Scores 10,000 LLM Responses on 0-4 Severity Scale
From Archives S&P 500 refuses entry for SpaceX, OpenAI and Anthropic, analysts surprised
Business & Startups
From Archives NSA uses Anthropic's Mythos AI model for offensive cyber ops against China, Iran
Open Source
Common Questions Answered
How did xAI continue accessing Claude after Anthropic revoked official access in January?
xAI engineers used personal accounts and a third-party service to log into Claude for months after Anthropic blocked their official access. This allowed them to continue training their models using Anthropic's technology despite the company's explicit decision to cut off their access.
What internal problems has xAI's model-building team experienced according to the article?
xAI's core team for building models fell below five people, with key leaders quitting the project. Additionally, someone accidentally deleted weeks of work, indicating systemic issues within the organization rather than isolated incidents.
What is the contradiction in Musk's current business strategy with xAI?
Musk is leasing xAI's own computing infrastructure to Anthropic and Google, the very companies he claims to be racing against in the AI competition. This creates a paradoxical situation where he is simultaneously competing with and supporting his rivals' operations.
What does the article mean by calling xAI's access to Claude 'data mooching'?
Data mooching refers to xAI's practice of continuing to use Anthropic's Claude service through unauthorized personal accounts after being officially denied access. The article uses this term to describe the unauthorized borrowing of a competitor's technology and data resources.
Further Reading
- Anthropic Revokes OpenAI's Access To Claude Over Terms of Service Violation — Slashdot / Wired report
- Anthropic blocks Claude's access to xAI Elon, AI giants start a war? — IDNFinancials
- Anthropic revokes OpenAI's access to Claude - Hacker News — Hacker News
- Today in AI: Anthropic Pulls Claude Code Access, Elon's xAI Team Caught Using Claude Internally — YouTube
- Detecting and countering misuse of AI: August 2025 — Anthropic
